21st of January 2015

By Professor Byron Sharp Director Ehrenberg-Bass Institute
Published by American Marketing Association

The Hole in the Leaky Bucket Theory

The leaky bucket theory suggests that companies are always losing customers, so to maintain share, you have to win an equal number of new customers to keep the bucket full, so to speak. To grow share, you have to be especially good at new customer acquisition, or you have to slow the leak.

The idea of plugging the leak became popular with theorists who sold the idea to practical marketers, who believed, without any evidence, that retention is cheaper than acquisition. The most extreme, and fanciful, example was from a 1990 Harvard Business Review article, “Zero Defections: Quality Comes to Services,” which alleged that the leak could be plugged for huge gains in profitability. Again, there was no empirical evidence.

In my book,  How Brands Grow, I emphasize a move toward customer acquisition. Many up-to-date marketers now accept the idea that gains in penetration are required for growth, so customer acquisition should be the emphasis of growth-oriented marketers. In terms of the leaky bucket theory, the emphasis has shifted from plugging the leak to accepting it. All brands lose customers, so the strategy is to work hard to fill the bucket with new customers at a faster rate than it leaks.

Research that we published last year in the Journal of Business Research supports this view. We show that brand growth and profits depend largely on a company’s ability to outperform similar-sized rivals in acquisition, while all brands suffer from close to expected rates of defection irrespective of whether they are growing or declining. However, I keep reading analyses that vastly overestimate how leaky the bucket is. They suggest that customers are astonishingly disloyal and are being buffeted around by marketing. They also suggest that massive recruitment efforts are needed just to maintain a customer base.

This started in 1996, when Joel Rubinson and Allan Baldinger of Port Washington, N.Y.-based NPD Group reported observing “incredible movement over time” and “much lower levels of year-to-year retention of high loyals than we expected.” Moreover, they said, “only 53% of high loyals to the brand remained highly loyal to the brand a year later.”

Such claims are still being made. For example, in 2013, Boston-based Bain & Co. reported that “on average, the top five brands in a category lose 30 to 60% of their shopper base every year.” Such analyses are misleading. The analysts have been tripped up by their data, and they pass this confusion on to their readers.

These discoveries of a horrendously leaking bucket are typically based on Big Data sets, recording the category purchasing of individuals or households over time (such as loyalty card members, or Nielsen and Kantar panel data). Analysts get a surprise when they calculate the proportion of customers who bought a brand in one period who return to re-purchase the brand in a following period. In consumer goods, only half of a brand’s customers—or fewer—come back. The proportion is better for the brand’s heavy customers, but it’s still alarming, and it’s horrifyingly low for people who did not buy frequently in the first period. The hole in the bucket looks to be very large, indeed.

Fortunately, things aren’t as bad as they seem. The law-like patterns documented in How Brands Grow are due to polygamous loyalty. If consumers are being buffeted about by marketing, it’s only at the margins, causing shuffling within their repertoire, but they maintain their loyalties over quite long periods. And these loyalties follow an NBD-Dirichlet distribution across consumers, which means, among other things, that the typical brand will have a great many light buyers.

It’s quite normal for a stationary brand (that is, one that’s not growing its customer base) with 5% penetration in a year to have 10% penetration over a three-year period because a very large proportion of its customers don’t buy it every year. If someone buys from you once every two years, then typically, if he appears in this year’s sales figures, he won’t appear in next year’s but will be back again the year after that. Such customers, and there are many of them, make the bucket look very much more leaky than it is. They also make your “recruitment” efforts look impressive.

In 1974, Andrew Ehrenberg wrote that there is no leaky bucket:

“The 55% who do not buy the brand in the second period are, however, not lost for good. Instead, they are merely relatively infrequent buyers of the brand who buy it regularly but not often. No special efforts have therefore to be made either to bring them back or to replace them (the ‘leaky bucket’ theory). Few things about the consumer in competitive markets can be more important than knowing this, and a successful theory of repeat buying was needed to establish it.”

Today, we have that successful theory, but Andrew wasn’t quite right: The bucket does leak a bit, although there is almost nothing that can be done about this small level of defection. And while he was technically right that no special efforts need to be made to ensure that most brand buyers maintain their loyalties, we do need to maintain the level of physical and mental availability that underpins our customers’ loyalty. Turning off advertising, or turning it in the wrong direction, won’t result in an immediate collapse in recruitment and, therefore, market share, but it will begin the seeping erosion of our market-based assets, triggering the inevitable decline in market share.


Now available as an eBook exclusively to Apple iBooks


The Ehrenberg-Bass Institute for Marketing Science is the world’s largest centre for research into marketing. Our team of market research experts can help you grow your brand and develop a culture of evidence-based marketing.